The propagation of uncertainties in stellar population synthesis modeling III: model calibration, comparison, and evaluation

Abstract

Stellar population synthesis (SPS) provides the link between the stellar and dust content of galaxies and their observed spectral energy distributions. In the present work we perform a comprehensive calibration of our own flexible SPS (FSPS) model against a suite of data. Several public SPS models are intercompared, including the models of Bruzual & Charlot (BC03), Maraston (M05) and FSPS. The relative strengths and weaknesses of these models are evaluated, with the following conclusions: 1) The FSPS and BC03 models compare favorably with MC data at all ages, whereas M05 colors are too red and the age-dependence is incorrect; 2) All models yield similar optical and near-IR colors for old metal-poor systems, and yet they all provide poor fits to the integrated J-K and V-K colors of both MW and M31 star clusters; 4) All models predict ugr colors too red, D4000 strengths too strong and Hdelta strengths too weak compared to massive red sequence galaxies, under the assumption that such galaxies are composed solely of old metal-rich stars; 5) FSPS and, to a lesser extent, BC03 can reproduce the optical and near-IR colors of post-starburst galaxies, while M05 cannot. Reasons for these discrepancies are explored. The failure at predicting the ugr colors, D4000, and Hdelta strengths can be explained by some combination of a minority population of metal-poor stars, young stars, blue straggler and/or blue horizontal branch stars, but not by appealing to inadequacies in either theoretical stellar atmospheres or canonical evolutionary phases (e.g., the main sequence turn-off). We emphasize that due to a lack of calibrating star cluster data in regions of the metallicity-age plane relevant for galaxies, all of these models continue to suffer from serious uncertainties that are difficult to quantify. (ABRIDGED)